Legacy and De Novo Banking Systems Flashcards

Legacy and De Novo Banking Systems

Banking is one of the oldest global industries, historically relying on technological advancements to facilitate operations. This evolution is characterized by the transition from legacy systems to de novo systems.

Legacy Systems

Computerization in banking gained momentum in the 1960s with the introduction of mainframe accounting systems. Today, these are referred to as "legacy" systems. They are typically decades-old and require significant upgrades.

• Challenges: Updating legacy systems is difficult because they are mission-critical to daily functioning. In the short term, banks often choose to "patch up" gaps as it is cheaper and faster.

• Long-term Risks: Persistent reliance on outdated systems leads to system outages. Older "incumbent" banks risk falling behind competitors due to cumbersome, inflexible systems.

De Novo Systems

Financial technology (fintech) companies utilize novel, web-based technologies known as de novo systems to deliver services. Unlike incumbent banks, these companies are not hindered by legacy infrastructure.

Historical Technological Influence

• 19th and 20th Centuries: Technology like railways, telephones, and computers allowed banking to expand beyond regional and national borders through increased efficiency.

• Modern Era: The internet, smartphones, and Artificial Intelligence (AI) current reduce the cost of distributing financial services.

Example: Evolution of Voice-Based Banking

1. Face-to-Face Era: Banking was originally a personal business where transactions occurred in person. Written agreements were often omitted in favor of the motto "my word is my bond."

2. Long-Distance Transmission: The telegraph enabled long-distance information sharing, such as trading shares between London and New York.

3. The Telephone: Invented in 1870, adoption in banking was slow. In 1989, Midland Bank (now HSBC) launched First Direct, the world's first telephone-only bank. It still provides round-the-clock service via a staffed call center.

4. AI Integration: Modern call centers use AI-based telephone bots as a first point of contact. These bots detect keywords (e.g., "mortgage") to route customers to specialists. This automation saves staff time for complex tasks.

5. Natural Language Processing (NLP): This technology enables digital voice assistants like Amazon Echo (Alexa) or Apple Siri. Beyond routine tasks, these assistants allow customers to check balances and make payments.

• The Trade-off: Consumers must accept a trade-off between cost and customer service (staff availability).

Key Types of Banking Institutions and Core Functions

Banking institutions vary by size and structure, primarily falling into three categories:

Types of Institutions

• Retail Banks: They take deposits from businesses and households (savers/depositors) and lend that money back to businesses and households (borrowers). They rely on branch networks and increasing digitization for efficiency.

• Building Societies: These are customer-owned institutions focusing primarily on home finance. Theoretically, they can offer higher interest rates to depositors and lower rates to borrowers. They have generally been slower to digitize due to localized geographic focus.

• Investment Banks (Wholesale Banks): These help large businesses or wealthy individuals access financial markets. They assist in issuing and selling securities. Due to international operations and demanding clients, they are the most heavily digitized incumbent group.

Core Banking Functions

Regardless of type, banks connect those with a surplus of money to those with a deficit. The five core functions are:

1. Taking Deposits: This process is called financial intermediation. The bank owes this money to the depositor.

   • Interest Margin: The difference between the rate paid to savers and the rate charged to borrowers.

   • Access: Short-notice funds offer lower interest; fixed-term savings offer higher rates.

2. Making Loans: Banks lend funds for various periods.

   • Terms: Overdrafts are repayable on demand; personal loans ($1\text{--}10$ years); mortgages (up to $35$ years or more).

   • Maturity Transformation: Converting short-term deposits into long-term loans. This is the foundation of banking.

   • Risks: Banks must manage liquidity risk (mass withdrawals) and credit risk (default).

3. Processing Payments: Banks are the backbone of the economy, processing cash, cheques, standing orders, direct debits, electronic transfers, cards, and mobile wallets.

4. Managing Assets: Two key sub-functions:

   • Asset Managers: Subsidiary companies helping customers invest in non-cash assets (stocks/bonds). The bank earns fees, and the client bears the risk.

   • Custodians/Depositaries: Storing physical assets (art, metals, deeds) or keeping digital ownership records of financial assets.

5. Providing Financial Advice: Selling insurance and pension products, often via "white labelling" (third-party products sold under the bank's brand). Specialist services for the wealthy (tax/estate planning) are known as private banking.

Legacy Infrastructure vs. Modern Processes

Legacy Infrastructure

Includes outdated installations like under-utilized branches and decades-old computer systems. Many large banks possess multiple legacy systems due to acquiring smaller rivals. Because they are mission-critical, upgrading is high-risk, leading to the use of "workarounds."

• Branch Decline: Retail UK banks are closing branches at a rate of $7\text{--}9\% \, \text{per year}$ due to digital shifts.

Legacy Processes

Some institutions still use paper-based back-office processes designed for branch-centric models.

• Evolution: Lending decisions used to rely on human relationship managers. Now, automated credit scoring systems using quantitative data are the norm for mass-market products (credit cards, personal loans).

• Specialization: Corporate lending often still requires human judgment and industry expertise.

Understanding Algorithms, Big Data, and AI

Key Definitions

• Algorithm: A decision-making process or set of tasks performed by a computer conditional on specific inputs.

• Big Data: Quantitative decision-making using extremely large datasets (text, audio, video, images). All AI is a form of big data, but not all big data is AI.

• Machine Learning: Algorithms that sift through datasets to detect patterns. They link variables (e.g., "smoking") to outcomes (e.g., "heart disease"), using hindsight to construct rules for future predictions.

The Definition of AI

While there is no single agreed-upon definition, the Turing Test (Alan Turing, 1950) suggests AI is when a machine's task result is at least as good as a human's. However, this bar is considered low (common objects like coffee makers might pass).

AI Subsets

• Neural Networks: Algorithms mimicking human brain cells. A basic network has two layers of nodes (connection points). The first layer assigns mathematical weights to variables; the second (decision) node applies the algorithm. Example: Google's search algorithm.

• Deep Learning: A neural network with multiple layers (up to $50$) between input and decision. This increases accuracy but makes the process opaque ("deep"). Many consider this the only "true" AI.

• Natural Language Processing (NLP): Algorithms that interpret human language (text/audio) based on a probabilistic or "best match" basis against dictionaries. These underpin chatbots.

AI-Led Evolution: Legacy Way vs. AI Way

Technology Enablers: Cloud and APIs

Cloud Computing

Historically, supercomputers were prohibitively expensive. The "cloud" leverages remote processing power accessed via the internet.

• Utility Model: Computing power is now a utility, paid for on a "pay-as-you-go" basis.

• Efficiency: Estimated to reduce electricity use by $80\text{--}90\%$. Providers (Microsoft, Google, Amazon) often house data centers in cooler climates to save energy.

Application Programming Interfaces (APIs)

APIs are programs governing interactions between different computer systems. They make data transfer safer and standardized.

• Banking Use: Allows banks to offer third-party services (e.g., share dealing from an external broker) directly through the bank's portal.

Financial Innovators and Mobile Banking

Types of Innovators

• Challenger Banks: Younger institutions with novel brands and smaller branch networks. Less burdened by legacy code. Often price products attractively to gain market share.

• Neo-banks: Technology-led, often mobile-only (e.g., Revolut, Monzo, Starling). No physical branches means lower costs. They use digital interfaces to connect with third-party providers (insurers/lenders).

• Peer-to-Peer (P2P) Lenders: Non-bank institutions often called "shadow banks." They distribute credit risk directly to savers rather than lending in their own name. ZOPA (the world's first P2P lender) uses machine learning to model credit risk and earns fees instead of interest.

• Robo-advisors: Provide automated, personalized investment guidance via apps. They use quizzes to determine risk tolerance and typically charge a percentage fee (e.g., $0.25\% \, \text{pa}$).

• Payment Apps: Include digital wallets (Apple/Google Pay) and budgeting apps. They collect data to design personalized products.

Mobile Banking Statistics

• 2011: Royal Bank of Scotland (RBS) launched the world's first mobile banking app.

• UK Adoption: Over $23\%$ of UK adults have a digital-only account; another $10\%$ intend to open one.

• Global Trends: Mobile-led banks are rising rapidly in Asia (Rakuten Bank, Line Bank, Digibank).

Consumer Responses to Digitization

Consumer preferences are shaped by experiences in other sectors (Netflix, Amazon, Uber). Banks must adopt strategies to meet these expectations:

• Omnichannel: Customers expect a seamless transition between channels (branches, apps, call centers).

• Human Touch: Digital fatigue is real; research shows human interaction builds higher loyalty than pure automation. Banks may use "credible AI imitations" to mimic this.

• Pricing: Increased price sensitivity due to rising costs and comparison websites.

• Convenience: Banks must remove "mental transaction costs" (uncertainty and wasted effort).

• Trust: Hard to quantify; requires transparency and ethical culture.

• Sustainability: $3 \, \text{in} \, 4$ UK adults are concerned about climate change; they favor banks with sustainability credentials.

• Personalization: Digital tech brings high-net-worth style service to the mass market via data analysis.

Technology Acceptance and Adoption

• Technology Acceptance Model (TAM - Fred Davis, 1989): Identifies factors for adoption: Personal attitude, task relevance, time/habit, perceived ease of use, and output quality.

• Technology Adoption Curve (Rogers, 1962): Categorizes consumers as Innovators, Early Adopters, Early Majority, Late Majority, and Laggards.

Generational Differences

• Strategic Implications: Wealth accumulates with age; Boomers need estate planning and retirement products, while Gen Z needs wealth accumulation and mortgages. Vulnerable customers (dementia/limited mobility) require branch-based human intervention.

Bank Responses and Innovation Tools

• Digitization Potential: McKinsey estimates $71\%$ of finance business can be transformed by technology.

• IT Spending: IMF estimates US bank IT spending in $2021$ was $6\times$ higher than in $2001$.

SCAMPER Tool for Innovation

• Substitute: Partner with fintechs (e.g., banks partnering with Wise for FX).

• Combine: The offset mortgage (Woolwich Building Society/Barclays) combining savings and debt.

• Adapt: First Direct adapting to a branchless model while keeping human phone support.

• Modify: Supermarket banks (UK) modifying loyalty card data (e.g., scratchcard spend vs. vegetable purchases) to assess credit risk.

• Put to another use: Lloyds Bank turning branches into co-working spaces with cafés.

• Eliminate: Starling Bank eliminating in-house data centers by using AWS (Amazon Web Services).

• Reverse/Re-organise: Tandem Bank starting as a budgeting app then acquiring Harrods Bank for its license.

Key Trends in Digital Evolution

Cashless Payments

Driven by digitization and "digital natives."

• Risks: Approximately $20\%$ of the UK population lacks basic digital skills. In Malaysia, only $19\%$ feel they have adequate digital skills for work. Privacy is also a concern regarding cash.

Big Tech

Companies like Amazon, Apple, and Google possess large customer bases and sophisticated AI data processing.

• Threat: They could offer automated lending more accurately than banks.

• Counter-Strategy: Banks should exploit Big Tech's over-reliance on automation by deploying a human touch.

Emerging Technologies

• Embedded Finance: Placing financial services into non-financial products (e.g., Buy-Now-Pay-Later/BNPL for a sofa).

• Micro-transactions: Facilitating tiny trades that were previously not worth the effort (Szabo, 1999).

• Virtual and Augmented Reality: AI-generated digital environments. Augmented reality enhances existing environments, while Virtual Reality can create fictional settings (e.g., a pension meeting on a virtual beach).